World's Best Driverless Metro Lines 2017

Home , Expo Line (SkyTrain), Metro (franchise), SelTrac, Shinbundang Line, Turin Metro

WORLD’S BEST DRIVERLESS METRO LINES 2017 Market study on driverless metro lines and benchmark of network performance www.wavestone.com

Wavestone is a consulting firm, created from the merger of Solucom and Kurt Salmon’s European Business (excluding retails and consumer goods outside of France) Wavestone’s mission is to enlighten and guide their clients in the most critical decisions, drawing on functional, sectoral and technological expertise.

2 Foreword

In tomorrow’s megacities, citizens’ megacities which are coming into view selected route will take on increasing in emerging countries in Asia, Africa and importance (in France, journey length South America as well as the challenges grew by 63% between 1982 and 2008 presented by the peripheral urbanization according to INSEE, France’s National of highly dense big cities in developed Institute of Statistics and Economic European countries. Studies). This panorama on “smartization”, which At the same time, citizens’ habits regard- optimizes and streamlines urban mobil- ing transport change as a result of pres- ity highlights France as the flagship of sure, from environmental responsibility driverless metro system operations. The which is more present in their conscience momentum of its authorities and industry and, on the other hand, from congestion in the segment has propelled the country in city centers. The “transport mix” in big to the top of the pack in the global driv- cities has clearly shifted from the individ- erless metro market. ual car towards mass public transport. A transport system’s performance is based Faced with the challenge of transporting on strategic choices made over the long more passengers in a continuous and fluid term by the organizing transport authority way, and with the challenge of increasing and tactical and operating choices made line capacity that is already saturated, the by the operator. driverless metro system brings practical solutions: high headway (up to 60’’ for Through its vision, the organizing trans- Lille’s metro), high flexibility (adapting the port authority comes up with a transport service frequency to real time demand), solution that meets mobility demands high service level (24/7 service, e.g. modelled for several years (generally, the Copenhagen’s metro) and higher commer- organizing transport authority’s vision is cial speed (up to 44km/h in Vancouver). for a 10-30-year period). In reality, this choice which seems to have little impact In addition, driverless metro systems con- on metro line operations proves decisive siderably reduce operating costs (by up to for: 40%) mainly bringing with them human resources flexibility. This means public // Mobility quality perceived by the transport prices are significantly reduced user, closely linked to the density and accessible to more of the population. of the stations (varying between 0.54 station/km for Dubai and 2.17 By making the most of all these advan- stations/km for Lausanne), with tages, the driverless metro system will, the theoretical headway and there- in the next five years, strategically meet fore the transport system’s capac- the challenges of decongestion in the ity (varying between 1’ for the Lille

*See glossary.

3 metro and 6’ for the Dubai metro), Transport service quality perceived by at the commercial speed brought the customer is also thought to be linked about by the choice of rolling stock to settings that are endogenous to oper- and accessibility, often standardized ations, which do not depend on the orga- by strict rules nizing transport authority. This perception // A recovery in the operator’s opera- of quality is achieved through a passen- tions which depend on the quality ger satisfaction level measured by the of the return good disposed of by operator itself, with the operator’s own the organizing transport authority indicators undergoing self-monitoring and the ageing of the network and i.e. monitoring by the organizing author- rolling stock. As a result, investment ity which seeks to measure the differ- for the operator and additional oper- ence between the thresholds set in the ating costs for maintenance are to be operations contract and users’ perceived planned for performance.

The study shows that the operator has a Wavestone chose to compare users’ satis- much more important role to play than the faction levels in two domains for which the organizing transport authority in improv- operator’s margins are significant: peo- ing the service delivered to the customer: ple’s cleanliness and safety. For the first the transport experience. It may compen- domain, lines including Lyon, Barcelona, sate for choices made by the organizing Rennes and Taipei top the rankings with transport authority which are judged as positive opinions from users, reaching sub-optimal, ensuring a high rate of punc- 99.3% for Lyon, versus less than 70% tuality and frequency (varying between for Vancouver and Milan. For the second 93% for Lausanne and 99.73% for Taipei) domain, the operator’s efforts are cru- and a group of innovative connected ser- cial even if the socio-economic context vices which transform urban mobility. of the area served by the metro line

Likewise, the operator’s expertise is cru- has an important role. Regarding safety, cial in guaranteeing efficient, reliable and Copenhagen, Dubai and Taipei come out above all profitable operations, particu- with user satisfaction rates that exceed larly for ageing networks and rolling stock 90%. such as that found on the following metro Nevertheless, the comparison of users’ systems: Lille (35 years’ old), Lyon (26), satisfaction between different networks Taipei (21) and Vancouver (31). Network remains a tricky task due: operations of this type may easily generate uncontrollable maintenance costs // The difference in terms ofmethods which have a direct impact on the orga- for measurement between the differ- nizing transport authority and the user ent lines and the type of surveys or through a higher transport price. inquiries conducted

4 // The diverse nature of the socio-eco- Finally, in its analysis, Wavestone adopts nomic differences of the served areas, a weighting factor which reflects the between cities and even countries old-age of the infrastructure and rolling // Citizens’ different perceptions faced stock. At similar operating performance, a with qualitative issues (safety, clean- network with major operating constraints liness) based on their country due to its age, reveals an operator’s potentially more advanced competency. // The scope of the inquiry which varies A change to this factor would have an according to author (operator, orga- impact on the global ranking: by reducing nizing transport authority and third such this factor, the best metro lines in the party). The scope for an authority or ranking may see their rank lowered, but third party is often larger than that the score will no longer precisely reflect of an operator, which is sometimes their operational excellence. impartial

As part of the development of its expertise in the trans- in the benchmark target as their context, ser- port sector, Wavestone’s Transport & Travel practice vice and operation are not comparable with has taken a deeper look at mass urban transport by “heavier” systems such as metros and are not drawing on the existing network of our international used in the same way. offices to enhance the knowledge base of our local and / The study analyses 25 of the 40 driverless metro international clients. lines existing worldwide in February 2017, for The trend of automation affecting all transport modes which information is considered as available, (including autonomous cars, autonomous buses and accessible and reliable, notably thanks to our hyperloops) quickly shifted the focus of the study to international offices, thus providing the items driverless metro lines, a mature and rapidly expanding necessary for a comparative analysis. technology, the data for which is fully accessible, unlike This survey is notably based on: those for disruptive systems, which are new and have / The collection of documents from different little market presence. sources and supported by Wavestone’s inter- To carry out the study, Wavestone limited the scope to national offices. a representative and coherent sample: / A series of interviews with major players in / Driverless metro lines* mainly transporting pas- urban transport and experts in the field. sengers on back-and-forth journeys including / Analysis work serving to compare driverless commutes. metro lines in three respects: the performance / Collective driverless transport lines such as light of infrastructure and rolling stock; the reliability rail transit (LRT)*, people movers* and auto- and quality of the service; and the innovation mated guided transit (AGT)* are not included demonstrated by the operators.

*See glossary.

5 Authors

before pursuing it in transport and multimodal Philippe Menesplier mobility consulting. Generalist of passengers and fret transport, expert Philippe Menesplier, Partner, in externalities and LCA Life Cycle Assessment, he has more than 20 years Juliano works regularly on strategic benchmarks of experience in strategy and studies in the service of major industry players: and management consult- operators, transport organization authorities and ing. He managed numerous equipment manufacturers. studies and clients’ deliveries for urban mobility market [email protected] actors: organization, industrial strategy, optimization and lean management. Also, he delivered important transformation pro- grams for companies evolving within the transport infrastructure industry, and he regularly conducts Jonathan Longo studies on multimodal mobility in megacities and Jonathan is part of Energy, agglomerations. Utilities and Transport [email protected] Practice of Wavestone. Following several years of experience in the energy sector (investment fund, bank, energy producer and distributor), he got the opportunity to help many Aurélien Gué private and public entities in the scope of market Aurélien Gué is a Senior analysis, business model conception and bench- Manager and expert in new marks deliveries. uses that are deeply chang- More specifically, he addresses problems mixing ing our day to day living and energy and transport: smart city, green mobility, the business model of the etc. industry players. [email protected] He managed several strategic studies related to new mobility solutions (air transport, mass-transit, etc.) and conducted major transformation projects in the fields of passenger’s experience and operations for major transport and Franck Benoit retail actors. Franck Benoit, Business [email protected] Analyst, he initiated his career in transport and travel consulting. He grad- uated from the “Institut d’Etudes Politiques d’Aix- Juliano Naoufal en-Provence”, public affairs Juliano Naoufal is a con- and holds a Master’s degree in Business Law from sultant within the Energy, Aix-Marseille University. Utilities and Transport His double expertise allows him to advise public Practice. Following several transport operators in conducting their strategic years of studies in transport projects. He is particularly interested in the rela- and sustainable develop- tion between mobility operators or providers and ment at the “Ecole des Ponts ParisTech”, Juliano transport organization authorities. started his career in the air transport industry [email protected]

6 Contents

09 Introduction of the study

29 Description of scope and methodology

Summary of the comparative analysis of driverless 35 metro lines

A detailed benchmark of the driverless metro lines 57 under reviews

Appendices 97

7 UN VISUEL

8 Introduction of the study

9 Mobility, a key issue in tomorrow’s cities

Urbanization is on the rise in in China and five in India. The figure will rise all the world’s megacities, to 41 by 2030. The world is currently home particularly in emerging to 7.3 billion people, over 54.5% of whom live in cities. By 2050, the world population countries will rise to roughly 9.7 billion, with over 65% of the total living in cities. The global urban landscape is in the midst of e deep-seated change. In 1950, only The number of megacities is on the increase, New York and Tokyo had a population of a trend accompanied by urban sprawl, over ten million. In 2016, some 31 cities are which gives these cities an interface role home to over ten million people, six of them through an effective transport system:

2016 - The World's 10 Biggest Cities (population in millions of inhabitants)

Tokyo 38 • China (19% of the world population) and Dehli 26.4 India (18%) are the world’s most populous Shanghai 24.5 countries, each with over one billion Bombay 21.4 inhabitants. São Paolo 21.3 • Only four megacities are not on the Asian Mexico 21.2 continent: São Paulo, Mexico, Cairo and Pékin 21.2 New York. Osaka 20.3 • None of the megacities are in Europe, and Cairo 19.1 Cairo is the only one in Africa. New York 18.6

2030 - The World's 10 Biggest Cities (population in millions of inhabitants)Chine France Allemagne

Tokyo 37.2 • Seven of the ten most populousItalie citiesRoyaume-Uni are États-Unis Dehli 36.1 in Asia Shanghai 30.8 • India will overtake China as the world’s Bombay 27.8 most populous country in 2022. Pékin 27.7 • Bangladesh (Dhaka) and Pakistan (Karachi) Dhaka 27.4 are highly populated and capital-centric. Karachi 24.8 • The population of Nigeria (Lagos) will Cairo 24.5 overtake that of the USA in 2050, with the Lagos 24.2 country becoming the world’s third most Mexico 23.9 populous. Source: United Nations, The World’s Cities in 2016. *See glossary. Chine France Allemagne

10 Italie Royaume-Uni États-Unis Africa and Asia are set for Asia (55%) and 3 in Africa (10%). In 2030, strong urban growth, leading 23 of the world’s 41 cities with over 10 to the advent of new megacities million inhabitants will be in Asia (56%) and 6 in Africa (15%). on the two continents The 10 new megacities will be in Asia Megacities are concentrated in Asia and (Pakistan, India, Thailand, Vietnam), Africa: in 2016, 17 of the world’s 31 cities Africa (South Africa, Tanzania, Angola) with over 10 million inhabitants were in and South America (Colombia).

Top 10 Cities with the Strongest Population Growth between 2016 and 2030 Hyderabad 38.6 Hyderabad 38.6 Asia (Bangladesh, Pakistan, India) Ahmedabad 39 Asia (Bangladesh, Pakistan, India) Ahmedabad 39 Bangalore 41.2 Africa (Tanzania, Angola, Nigeria, RDC) Bangalore 41.2 Africa (Tanzania, Angola, Nigeria, RDC) Lahore 45 Lahore 45 Karachi 45.1 Karachi 45.1 Dhaka 50.1 Dhaka 50.1 Kinshasa 65.7 Kinshasa 65.7 Lagos 77.4 Lagos 77.4 Luanda 81.8 Luanda 81.8 Dar es Salaam 98.9 Dar es Salaam 98.9

Geographical Breakdown of Cities Entering the Rankings of Megacities with Over 10 million Inhabitants in 2030

10 10 Chine France Allemagne Chine France Allemagne Asia Italie Royaume-Uni États-Unis Asia Italie Royaume-Uni États-Unis 30 60 Africa 30 60 Africa South America South America

Source: United Nations, The World’s Cities in 2016.

2000 The rapid expansion of these cities, and the emergence of a middle class resulting from the far-reaching1500 economic transformation in these regions, make urban mobility a key issue for these countries. The urban transport market will be increasingly concentrated in these two regions in the coming decades. 1000

500 500 11

0 To prepare for double-digit Mobility that goes above urban growth, the world’s and beyond the tradi- biggest cities need to start tional, technical and functional definition of transport thinking about sustainable to adopt a service-based and more mobility here and now modern approach: mobility conditions (comfort and well-being, wait- Worldwide, the urban population overtook ing times, information access) are the rural population in 2007. In a world of becoming crucial, with users more cities, city dwellers, whether commuting concerned about the time than the or traveling for family or leisure purposes, distance of their daily journeys. are in movement: mobility is an integral part of their everyday lives. Mobility that takes account of the back-and-forth Today’s cities are congested, traf- travel characteristic of fic-jammed and polluted. Tomorrow’s large cities, in which inhabitants take cities need to address the mobility issues the same routes back and forth in of their inhabitants – and the manage- the same day, mainly as part of their ment of transport flows has become a commutes. crucial issue. Tomorrow’s hyper-connected and smart cities (which are just around the corner) also need to take a new approach Clickable, efficient and to mobility: personalized transport for Mobility adapted to the augmented mobility in the Smart characteristics of tomor- City row’s cities: the modern metropolis – the “Smart City” – is For economic and ecological reasons, larger and more densely populated, large cities around the world are seeking a place of 24-7 mobility with peaks in to shift to a Smart City model. They are transport use that require transport exploring a broad range of fields – includ- authorities to be adaptable. ing energy, waste management, water supply and enhanced transport – and Mobility that respects the digital technology is playing a vital role in : mobility environment optimizing the way these services work. today is overly polluting; it needs to be approached as part And so the transport sector is in the of the efforts to shrink the environ- midst of a major transformation, with mental footprint of the Smart City. far-reaching changes in the offer and in

12 user behavior since the arrival of digital is supplied to transport authorities technology. A day-to-day issue for all the through partnerships (as with the population, transport is naturally a key Waze Connected Citizens Program). focus in efforts to develop Smart Cities.

Data is used to assist passengers, who now have Shorter journeys and lighter access to an unprece- transport modes around dense dented urban transport offer with and increasingly decentralized increasingly varied transport modes agglomerations (collaborative modes, bike fleets, self-service cars, etc.) accessible in Commutes are growing increasingly just a few clicks thanks to digital longer. The average commute distance in advances in the sector. Transport France has risen 63% in the last 30 years. data, widely shared on an open-data This is mainly a result of the widespread basis, are behind new comparative migration of inhabitants to peri-urban services, such as CityMapper. areas, which in turn results from the sub- stantial growth in the population of cities The availability of trans- and increasing pressure on urban areas. port data is also enabling transport organization The socio-economic and ecological authorities to better plan the impact of the daily commutes stem- development of their networks by ming from this shift is considerable. The adapting transport offers to pas- economically active population is more senger flows, with electronic tickets stressed and productivity is on a non-stop improving the analysis of journeys, decline, negatively impacting the urban connections and so on. economic fabric.

The use of real-time To address this problem, countries, and sharing is improving often developed countries, are seeking the responsiveness of to bring homes closer to workplaces and the transport offer and serving to even concentrate them in peri-urban enhance mobility. Transport users micro-centers (such as those in Seine are increasingly called on to provide Saint-Denis and Paris-Saclay outside information on the functioning of Paris). the network in real time (with Waze and Moovit, for example), and the The adoption of this approach in the information collected in real time national strategies of various countries,

13 along with the rise of these “micro-cities”, Average Commute Distance in France whose populations are considerably lower (in km) than those of cities, presages a switch from a mass transit network, such as the 15 metro, to much lighter micro-networks +63% such as people movers and trams. 10 A product of this model, the Grand Paris Express rapid transit system serves to decentralize the City of Paris and paves 5 the way for the arrival of tomorrow’s peri-urban micro-cities.

0 INSEE. Source: 1982 1994 2008

Current Urban Transport Model Future Peri-Urban Transport Model

Metro-style mass transit line Light tram or people-mover line with very low capacity

14 Tomorrow’s smart city will be Copenhagen and Toulouse to Dubai served by a higher-performance and Vancouver; mass transit solution: the > it is more flexible to operate, both in technical and HR terms: automatic driverless metro solutions can be quickly adapted to “New mobilities” are a new way for get- available capacity by minimizing or ting from A to B made possible by emerg- even eliminating constraints linked ing models and technologies and rolled to drivers. out through high user adoption rates, It creates a better service including electric cars, autonomous shut- for passengers tles and carpooling. But these systems > it helps to reduce wait- are not adapted to mass transit, involving times on platforms by providing faster passenger throughput rates on ing a more frequent service and high-pressure routes and in a sustainable faster commercial speed than con- manner. ventional metros while maintain- Driverless metros stand as an intelligent ing top-level punctuality; and innovative mass-transit solution. > it offers a better passenger experi- Driverless technology meets a certain ence through more recent trains and number of the objectives involved, includ- renovated stations. ing high capacity, speed and regularity, It is a reliable means of reduced operating costs, adaptability, and transport requiring lead- flexibility in terms of human resources. It ing-edge expertise fulfills the criteria of a new approach to By eliminating the risk factors mobility: stemming from human driving, the It is adapted to the driverless metro achieves higher requirements of future safety and reliability rates than cities conventional metro systems*; the > it is integrated in the urban envi- introduction of platform doors ronment and responds to the also limits the risk of accidents and demand for diversified mobility from human presence on the track.

*See glossary.

15 For example, the driverless selected with a view to decentralizing the metro is central to the Greater city, reducing commute times and, above all, Paris transport strategy and creating peri-urban socio-economic centers of interest able to relieve congestion in the the decentralization of the city capital.

The initial part of the Greater Paris trans- The law relative to Greater Paris provides port plan concerns the modernization for a specific organizational model for the and extension of the existing network. Grand Paris Express, whereby technical This includes extending the RER E line and infrastructure management is handled by the metro line, creating high-level-service RATP and the operation of the lines is han- buses and trams, modernizing RER trains dled by operators selected through tenders. and improving Transilien regional train lines. While the reform of collective transport in The second part of the transport plan is Île-de-France brings all operators access focused on the creation of new driverless to the network in free and non-discrimina- (and formally provides for metro lines, the “Grand Paris Express”. The tory conditions a seamless connection between the activ- aim is to build 205km of metro lines and 72 ities handled by RATP and those handled new stations, to enter into service between by the network manager), operators have 2018 and 2030. expressed reserves as to the organization As with similar projects in other large model selected and the fairness of the call cities, the Grand Paris Express project was for competition.

16 17 Key figures on driverless metro lines worldwide and the financial performance of automation

The world’s 40 driverless metro* lines are concentrated in the EMEA region and South-East Asia

Nearly 3/4 of driverless metro lines are 81% of driverless metro lines are located in located in cities with dense urban popula- cities with over 500,000 inhabitants. tions (over 4,000 inhabitants/km²).

Worldwide 40 Lines in 26 Networks NB: The UITP Observatory of automated metros has counted 55 lines due to including in its scope certain people EMEA: APAC: 25% mover lines, LRT systems, AGT systems, trains with a >100 63% Asia 25 passagers/train capacity (vs a capacity of >400 PAX/train for Europe the Wavestone study), monorail trains (vs only due or 3-rails South America for the Wavestone study) as well as Maglev technology (vs 57,5 2,5 30 AMERICAS: only pneumatic or steel for the Wavestone study): scope 10 13% differences with this benchmark Observatory attributable 5 North America to the fact that the Wavestone study aims to compare performance on equivalent urban mass-transit technologies Middle East and use.

*In the scope of this study, the definition of “driverless metro” is slightly different from the one adopted by the UITP, that considers 55 lines. The UITP includes in its definition people movers, LRT and AGT, which is not the case of the current benchmark.

18 List of Driverless Metro Lines Worldwide Copenhagen M1 Paris Over 5m inhabitants (city population) M2 The lines analyzed in this study are underlined Line 1 Lille Between 1m and 4.9m inhabitants (city population) Line 14 Line 1 London Under 1m inhabitants (city population) Nuremberg Docklands Light Railway Line 2 Over 10,000 inhab./km² (population density) Line U2 / Line U21 Between 5,000 and 9,999 inhab./km² (population density) Line 3 Rennes Under 5,000 inhab./km² (population density) Line A Budapest Lausanne M4 Lyon M2 Line D Toulouse Brescia Line A Turin MetroBs Line 1 Vancouver Line B Canada Line Rome Skytrain Expo Line Milan Line C Skytrain Millenium Line Barcelona Line 5 Line 9 Toronto Line 10 Scarborough RT Line 3 Line 11 Dubai Red Line Green Line Seoul Shinbundang Line Uijeongbu - U Line

Shanghai Busan São Paulo Line 10 Line 4 Line 4 Yellow Taipei Hong Kong Brown Line South Island Line

Kuala Lumpur Sungai Buloh-Kajang Line Singapore Circle Line (CCL) North-East Line (NEL) Downtown MRT Line (DTL) Copenhagen M1 Paris Over 5m inhabitants (city population) M2 Line 1 Lille Between 1m and 4.9m inhabitants (city population) Line 14 Line 1 London Under 1m inhabitants (city population) Nuremberg Docklands Light Railway Line 2 Over 10,000 inhab./km² (population density) Line U2 / Line U21 Between 5,000 and 9,999 inhab./km² (population density) Line 3 Rennes Under 5,000 inhab./km² (population density) Sources: Wavestone research and analysis, UN data, UITP. Line A Budapest Lausanne M4 Lyon M2 Line D 19 Toulouse Brescia Line A Turin MetroBs Line 1 Vancouver Line B Canada Line Rome Skytrain Expo Line Milan Line C Skytrain Millenium Line Barcelona Line 5 Line 9 Toronto Line 10 Scarborough RT Line 3 Line 11 Dubai Red Line Green Line Seoul Shinbundang Line Uijeongbu - U Line

Shanghai Busan São Paulo Line 10 Line 4 Line 4 Yellow Taipei Hong Kong Brown Line South Island Line

Kuala Lumpur Sungai Buloh-Kajang Line Singapore Circle Line (CCL) North-East Line (NEL) Downtown MRT Line (DTL) Overview and Key Figures of the World Driverless Metro Market

OPERATORS OPERATING THE GREATEST LINE MILEAGE FRENCH ENTITIES OPERATE NEARLY 1/4 OF THE WORLD'S 5 out of the 40 lines 104 NETWORKS (23.1%), FOLLOWED CLOSELY BY THE ITALIANS (19.2%) are operated by foreign players FRANCE ITALY CANADA SOUTH KOREA CHINA 75 68 51 41 41 87.5% of the lines 36 36 35 31 7% 28 26 23.1% 7% 7% are operated by 19.2% national operators

ATM TMB Keolis Serco Tisséo RATP Prasarana NeoTrans SBS Transit / SMRT Trains ComfortDelGro Translink / BCRTC OPERATORS OPERATING THE LARGEST NUMBER OF STATIONS SMRT Corporation / Shanghai Jiushi Group Shentong Metro Group / 137

TOP 5 COUNTRIES IN TERMS OF DRIVERLESS LINE MILEAGE 50 49 41 120 38 35 34 34 31 31 30 93.2 76.5 74.6 57.9 ATM TMB Keolis Serco Tisséo RATP Prasarana SBS Transit / SMRT Trains ComfortDelGro Translink / BCRTC SMRT Corporation / Shanghai Jiushi Group Shentong Metro Group / France Canada Singapore United Italy Arab Emirates THE MAJORITY OF DRIVERLESS METRO LINES ARE OPERATED BY PRIVATE EXTERNAL PLAYERS TOP 5 CITIES IN TERMS OF DRIVERLESS LINE MILEAGE 86.8 76.5 74.6 Managed by external player 40 Managed by local player 51 (private law) 60 (public service) 44

Vancouver is the only network with 1 line managed by an external player (SNC-Lavalin) Vancouver Singapore Dubai Kuala Lumpur Lille and the other 2 lines by a public player (municipal council).

20 OPERATORS OPERATING THE GREATEST LINE MILEAGE FRENCH ENTITIES OPERATE NEARLY 1/4 OF THE WORLD'S 5 out of the 40 lines 104 NETWORKS (23.1%), FOLLOWED CLOSELY BY THE ITALIANS (19.2%) are operated by foreign players FRANCE ITALY CANADA SOUTH KOREA CHINA 75 68 51 41 41 87.5% of the lines 36 36 35 31 7% 28 26 23.1% 7% 7% are operated by 19.2% national operators

Vancouver is the only network with 1 line managed by an external player (SNC-Lavalin) Vancouver Singapore Dubai Kuala Lumpur Lille and the other 2 lines by a public player (municipal council).

21 Driverless metro lines are // Station: the investments involved considerably more competitive in implementing platform doors are than conventional metros and offset bysmaller platform areas for driverless metros and, consequently, will become more so as they are lower civil engineering costs. This industrially developed results from the higher-level service of the driverless metro in terms of Highly competitive in terms of complete headways. cost: // Line: the costs of driverless lines are // Rolling stock: the additional cost naturally higher than those of conven- linked to the purchase of driverless tional lines. This is largely due to com- trains is offset bya gain in the sizing munication and signaling. However, of the rolling stock, since automatic the use of CBTC communication is systems have better performance becoming more widespread for differ- (headways / commercial speeds). For ent types of metro, thereby reducing the same service frequency, the driver- the cost gradient involved. less metro requires 33%* fewer trains.

Investment Costs per Km (in € million)

21 0.5 27 +12% 15 –40% 15 0.2 0.2 108 87 0.3 0.36

0.15 0.13 Classic metro Automatic metro Classic metro Automatic metro

Lines Stations Rolling stock Energy Maintenance Staff Staff on board

*Rolling stock manufacturers, Wavestone analysis.

22 // Operating costs: the additional cost linked to the maintenance of platform protection systems is offset by savings In terms of global costs per on personnel costs and energy costs (15% lower for driverless metro lines). km, the driverless metro is more competitive than the conventional metro for a depreciation of infrastructure and rolling stock of over 30 years.

Operating Costs per Km (in € million)

21 0.5 27 +12% 15 –40% 15 0.2 0.2 108 87 0.3 0.36

0.15 0.13 Classic metro Automatic metro Classic metro Automatic metro

Lines Stations Rolling stock Energy Maintenance Staff Staff on board

Hypotheses: Average line length of 14.5km, 30 trains per line, 22 stations on average per line, 200 drivers to operate a line. Sources: LVMT Laboratoire ville mobilité transport, The Economics of Urban Transportation 2007,

23 Technological trends and challenges in metro automation

Operators need to adopt their by Siemens in 2013), indicate the future know-how to the growing use of widespread use of the technology in CBTC technology the rail sector in general and driverless metros in particular. CBTC technology is the preferred sig- All the metro systems analyzed in this naling solution with a 68% share of the study are equipped with CBTC signaling market. This solution combines safety and and communication. A wide range of sup- optimized frequency. pliers are represented by the benchmark The market of CBTC solutions suppliers is (see graph opposite). highly concentrated: the top 3 suppliers, The Lille network is currently being namely Bombardier, Siemens and Thales, redesigned and the new system is to have a 78% share of the market (including be equipped with Urbalis, supplied by for non-automatic trains and metros). Alstom. But the delay in the project (esti- The convergence of manufacturers mated at 4 years) means that the current towards CBTC, as well as the invest- system (supplied by Siemens) will be ments involved (Invensys was bought out maintained for the next few years.

Market Share* of Communication Breakdown of CBTC Subcontractors in System Suppliers (in %) the Sample Studied (number of lines)

Bombardier 1 22 Siemens Taipei 40 Thales Alstom 1 Lausanne 19 Bombardier Other actors Ansaldo STS 2 20 Milan (1), Copenhagen (1)

Thales 6 *Frost & Sullivan data, Wavestone analysis. London (1), Dubai (2), Vancouver (3)

Siemens 12 Toulouse (2), Lille (2), Rennes (1), Paris (2), Turin (1), Barcelona (1), Nuremberg (2)

Sources: World Report on Metro Automation - July 2016, UITP. 24 The share of pneumatic tires for 70% of total mileage built. But pneu- is falling to the benefit of matic links have an advantage in terms of steel wheel-rail links due sound levels and passenger comfort. to the specific expertise and maintenance constraints involved The know-how of the operator and technological expertise are decisive. The two 10 manufacturers supplying rolling stock. technologies have their specific charac- Bombardier, Alstom and Siemens are the teristics in operation, with pneumatic leaders with 68% of total installed mileage. systems requiring know-how in maintenance. In this respect, some operators, The preferred wheel-rail link in the mileage such as Serco and ATM, appear to prefer built in the last decade is steel, accounting steel links.

Market Share of Steel vs. Pneumatic for Analyzed Sample (in %)

Toulouse London (DCL) Lille Copenhagen Rennes Milan Lyon (D) Vancouver Lausanne 48 52 Dubai Paris (1 & 14) Barcelona Turin Nuremberg Taipei

Pneumatic Steel

PneumatiC Steel

• Better grip, system adapted to cities with • Lower energy consumption owing to reduced + sharp gradients + rail friction • Less noise inside and better passenger • Better adapted to metro systems in cities comfort with difficult weather conditions (e.g. Dubai, • Better adapted to automated system by Nordic countries, etc.) avoiding wheel slip • Lower maintenance costs

• Not particularly energy efficient (+1% to 3%) • Recurring slip of train-rail link leading to – • Leads to higher maintenance costs – control system problems • Makes the system heavier, which means it • Longer headways owing to longer breaking consumes more energy and acceleration times • Higher fine-particle emissions than steel

25 What does automation involve // The possible creation of new main- for operators? tenance workshops adapted to solutions that are often specific to Technically speaking, operators need to: driverless metro lines, such as platform doors. // Carry out the prerequisite work for the installation of platform doors Operationally speaking, operators need necessary for optimizing passenger to: safety. // Ramp up the driverless rolling stock // Roll out the automation system for during the transition phase operating train operation, which differs depend- in combined mode. Operators have ing on the desired headways: to make this transition with as few • For long headways, operators interruptions as possible. The com- can opt for a fixed-block com- missioning phase increases risks in munication system (in which the service quality and requires specific position of the train is determined expertise on the part of the operator. relative to a section of the line). This is a mature technique used Organizationally speaking, operators by driverless and manual metro need to: lines. However, because it is only // Redeploy driving staff to con- moderately precise, it cannot be nected positions (line control and used for short headways. supervision).

• For short headways, operators // Upskill existing staff or hire the new are required to implement a employees required to ensure the mobile-block communication operation and maintenance of driv- system (in which train position erless lines. and speed are determined in a precise manner). This necessarily calls for the roll-out of additional technologies, namely ATP*, ATC* and ATO*. This in turn leads to additional risks (of a technical nature).

*See glossary.

26 The development outlook for driverless metro lines

The outlook for the development In 2025, Asia and Europe are expected of driverless metro lines looks to account for 33% and 30% of driverless robust through 2020, driven by metro mileage, followed by the Middle East (25%) with ambitious projects such large-scale projects as the Riyadh metro.

With more and more people living in cities, China has announced the introduction mass transit systems increasingly have to: of two new driverless lines between now // Optimize the existing urban and end-2017, one of them built exclusively infrastructure as capacity can be with Chinese technologies. extended only to a limited extent. The solution in this case is the automation Completed and Projected Driverless Metro of conventional lines. Lines* in Km and Growth Over 10 Years

// Develop new infrastructure on the 276% 2,000 outskirts of cities offering users 1,800 high-quality service while keeping costs under control. 1,600 1,400 Automated technologies propose solu- 1,200 tions meeting both these requirements. 1,000 According to the UITP, by 2025 some 800 2,300km of driverless metro lines will 600 143% be in operation, compared with around 400 800km today. 103% 200 The increasingly widespread use of CBTC 0 1990 2000 2010 2020* technology by all driverless metro lines together with rising living standards in Source: World Report on Metro Automation - July 2016, emerging countries will boost the com- UITP. petitiveness of automation solutions and *The estimated number of kilometers by 2020 takes account of construction and/or automation projects increase demand. confirmed as of July 2016.

27 28 Description of scope and methodology

29 Scope of the study

Data access and reliability enabled us to analyze 25 driverless metro lines* out of the total 40 existing in February 2017 within the scope of the study

Criteria Used to Define Scope

Driverless metro*: a dedicated, rail-powered circuit mainly used for commutes, with several carriages (up to 6 or 8), raised access and a capacity of over 400 passengers

Not including automated people-mover*, light rail transit* and automated guideway transit* systems

Pneumatic or steel rail link

Excluding Maglev and Hyperloop rail links

Dual-rail or third-rail systems

Excluding monorail

Automation level GoA 3 and 4*

Automation level GoA 1 and 2* excluded

*The terms used in the study are defined in the glossary.

30 List of Driverless Metro Lines Worldwide

Network included in the comparative analysis Network not included in the analysis due to the

unavailability or unreliability of information Copenhagen M1 Paris Over 5m inhabitants (city population) M2 Line 1 Lille Between 1m and 4.9m inhabitants (city population) Line 14 Line 1 Under 1m inhabitants (city population) London Nuremberg Docklands Light Railway Line 2 Over 10,000 inhab./km² (population density) Line U2 / Line U21 Between 5,000 and 9,999 inhab./km² (population density) Line 3 Under 5,000 inhab./km² (population density) Rennes Line A Budapest Lausanne M4 Lyon M2 Line D Toulouse Brescia Turin Line A MetroBs Line 1 Vancouver Line B Canada Line Rome Milan Line C Skytrain Expo Line Barcelona Line 5 Skytrain Millenium Line Line 9 Toronto Line 10 Scarborough RT Line 3 Line 11 Dubai Red Line Green Line Seoul Shinbundang Line Uijeongbu - U Line

Shanghai Busan São Paulo Line 10 Line 4 Line 4 Yellow Taipei Hong Kong Brown Line South Island Line

Kuala Lumpur Sungai Buloh-Kajang Line Singapore Circle Line (CCL) North-East Line (NEL) Downtown MRT Line (DTL)

31 Benchmark methodology

Performance Fields Selected for the Benchmark

Performance field

1 Performance of infrastructure and rolling stock

2 Service reliability and quality

2 Auxiliary or innovative services

The difference in performance between 2 driverless metro lines can be attributed to a great extent by the operator’s ability to effectively operate the lines in question. Infrastructure and rolling stock are key to the performance of a network but with room for maneuver in the much longer term.

32 Icons denoting the fields to which the analysis of each page refers

Indicators concerned Proposed weighting Commercial speed Distance between stations Headway 1 Accessibility Intrusion rate in network

Punctuality and/or availability and/or regularity Cleanliness Passenger information 1.5 Passenger safety

Available communication networks Field assessed but not included in the Innovative mobility service (route planner, etc.) final grade attributed New technologies (NFC, IoT, mobile) to each metro line

33 34 Summary of the comparative analysis of driverless metro lines

35 Perspective

The performance of a transport system – The architecture and geography including driverless metros – is impacted ** of the network: involving the by a set of factors external to the orga- uneven exposure of rolling stock nization of the operator or transport to its environment (underground, authority overground) and specific needs for fulfilling transport demand (air This means that the distinctive character- conditioning, heating). istics of each network have to be taken into account in order to contextualize the The “load curve” (breakdown of analysis ** passenger use over the day): involving peaks of varying extent Consequently, the following character- that may have an impact on punc- istics or factors need to be considered tuality (passenger incidents, etc.). when comparing performance: The socio-cultural characteristics The age of the network: entailing ** of the user population: involving * growing obsolescence costs with the uneven use of public services, the age of the infrastructure and with particular impact on cleanli- rolling stock, as well as the corre- ness and safety. sponding adapted processes for maintaining the line in operational condition.

The requirements included in the * delegation contract (headways): *Factor taken into account in the grade involving an operational strategy attributed to the metro line. on the part of the operator capa- ble of delivering performance **Factor not taken into account in the consistent with contractual grade, but assessed in the analysis of the requirements. results.

36 The performance of the world’s Grades for the Performance of driverless lines: a perfect Driverless Metro Lines (out of 5) balance between infrastructure Infrastructure x Operation / Network age and operation at the service of Lille - Line 2 4 passengers Lille - Line 1 4 London - Docklands 27 - 37 The performance of driverless metro lines Light Railway 4 is assessed through a complex equation of Toulouse - Line A 3.9 the operational ability to deliver a trans- Taipei - Wenhu Line 3.8 port service adapted to mobility behavior Lyon - Line D and the optimization of the transport con- 3.8 ditions of the journey experience. Rennes - Line A 3.8

// The correlation between the strong Paris - Line 14 3.7 constraints imposed by transport Vancouver - Expo Line 3.6 authorities (short headways) and the Toulouse - Line B 3.2 punctuality of the service is vital to Copenhagen - M1/M2 3.1 performance. Turin - Line 1 3.1 // Ageing networks require opera- Dubai - Red line tional overperformance on the part 3.1 of operators. Dubai - Green line 3

// The experience operators have of a Nuremberg U2 2.9 network and their length of service Nuremberg U3 2.8 are decisive to the performance of Paris - Line 1 2.7 an driverless metro line. Lausanne - M2 2.6 The design, construction and commis- Barcelona - Line 9 2.5 sioning of an driverless metro line present Milan - Line 5 2.4 value added for the city and its popula- Vancouver - Canada Line tion, as vital as it is a difficult activity for 2.3 transport authorities and operators. Vancouver - Millenium Line 2.1 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 The most recent driverless lines and networks demonstrate high levels of reliabil- ATM Keolis Serco TL BCRTC SNC Lav. TBM TRTC ity that need to be maintained over the GTT RATP Tisséo VAG long term if they are to become world leaders. Sources: Wavestone grades and modeling. The grade being the weighted sum of infrastructure performance (weight=1), operational performance (weight=1.5) and the network age indicator (weight=0.5).

37 The highest-performance driverless metro lines share the same characteristics of excellence

The best driverless metro transport services all score high on 4 key points related to infrastructure and operation:

The able management of Strong punctuality ageing infrastructure

These operators demonstrate excel- This is a defining characteristic of driv- lence in the management of old assets. erless metro lines. But combined with As confirmed by the impressive short headways it confirms theabil - availability and regularity of ity of an operator to respect their networks. the requirements imposed URE PU CT NC . RU TU by the transport authority T A S L A I R T F Y N I

H T EA R D O W MF AY CO High-level user satisfaction S Short headways relative to cleanliness and safety

The strong punctuality of the lines Globally speaking, the surveys led by managed by these operators confirms these operators confirma good level their commitment to demanding trans- of user satisfaction, knowing that port authorities. ageing infrastructure has a negative impact on passenger perception in this area.

38 Infrastructure and rolling stock performance

The performance of the Grades for the Performance of infrastructure and rolling Infrastructure and Rolling Stock* (out of 5) stock of the world’s driverless Vancouver - Expo Line 4.3 metro lines underpins their Toulouse - Line A safety and comfort 4.3 Lille - Line 2 4.2 29 - 39 Infrastructure and rolling stock perfor- Lille - Line 1 4.2 mance involves: Paris - Line 14 3.9 Scoring de performance // Strategies on urban mobility flow London - Docklands de l’infrastructure et du Light Railway 3.8 management: the design of a driv- matériel roulant* Taipei - Wenhu Line 3.8 erless metro line, from upstream research to the layout of the line and Lyon - Line D 3.7 the choice of stations, determines the Rennes - Line A 3.7 relevance of a line and its utilization Toulouse - Line B 3.5 success. Copenhagen - M1/M2 3.2 / an infrastructure / Network access: Turin - Line 1 3.2 that makes access easier for people Vancouver - Millenium Line 3.1 with mobility issues or harder for fraudulent users contributes to over- Nuremberg U2 3.0 all quality. Milan - Line 5 3.0

// Matching operating requirements Nuremberg U3 2.9 and expected use: the optimization Dubai - Red line 2.7 of infrastructure and rolling stock Dubai - Green line 2.6 hinges on aligning service levels with Barcelona - Line 9 2.6 use. Paris - Line 1 2.6 Infrastructure operation and maintenance Lausanne - M2 2.6 serve to maintain performance levels over the long term. This challenge tests the Vancouver - Canada Line 2.5 operational performance of operators. 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 ATM Keolis Serco TL BCRTC SNC Lav. TBM TRTC GTT RATP Tisséo VAG

*The sum of the grade attributed to the performance of the network and stock (weight=1) and the grade attributed to the aged of the infrastructure (weight=0.5) (5 corresponds to very old infrastructure).

39 Four main categories of infrastructure and rolling stock performance for driverless metro lines

efficient BEGINNER High performance/ High performance/ Old network Relatively new network

5 40 4.8 35 4.6 30 4.4 4.2 25 4 20 3.8 15 3.6 10

3.4 Age of infrastructure (in years) 3.2 5 3 0 Infrastructure and rolling stok performance (out of 5)

Vancouver - Millenium Line Lille - Line 1 Lille - Line 2 Milan - Line 5 Turin - Line 1 Lyon - Line D Paris - Line 1 Paris - Line 14 Nuremberg U2 Nuremberg U3 Lausanne - M2 Rennes - Line A Dubai - Red line Light Railway Toulouse - Line A Toulouse - Line B Barcelona - Line 9 Dubai - Green line Taipei - Wenhu Line London - Docklands Copenhagen - M1/M2 Vancouver - Expo Line Vancouver - Canada Line Vancouver - Millenium Line Performance infrastructure Age of infrastructure rolling stock

The players operate infrastructure The entities in this category operate and rolling stock that are generally infrastructure and rolling stock that recent and even new. Their strong are several years old. Their strong performance may be considered as performance reflects theiroptimal resulting from the newness of the use of these assets, which requires network. At this stage, the good good operating conditions on the performance of the rolling stock cannot part of the transporters. be correlated to any operator input.

40 AGILE STRUGGLING Average performance/ Low performance/ Very old network Relatively new network

5 40 4.8 35 4.6 30 4.4 4.2 25 4 20 3.8 15 3.6 10

3.4 Age of infrastructure (in years) 3.2 5 3 0 Infrastructure and rolling stok performance (out of 5)

The entities in this category have The input of operators in this difficulties maintaining their rolling category is clearly a factor in the stock in optimal operating condition, performance of the assets. The independently of the fact that their advanced age of the rolling stock stock is new. As such, the ageing suggests considerable maturity on of the infrastructure and the rolling the part of the operator in terms of stock will have a considerable impact maintenance and operations. on performance.

41 Line density is a factor in the Average Distance Between Stations (in m) reduction of travel times, the Lausanne - M2 improvement of the passenger Rennes - Line A experience and the optimization Nuremberg U3 of infrastructure Turin - Line 1 Driverless metro networks are character- Milan - Line 5 ized by between stations. short distances Paris - Line 1 European lines are denser in this respect Lille - Line 2 than international lines, mainly because Toulouse - Line A European cities have smaller surface Lille - Line 1 and public transport areas users prefer London - Docklands to walking. Light Railway Toulouse - Line B Adding more stations to a network or line Nuremberg U2 involves additional operating constraints (the more stations, the higher the punc- Lyon - Line D tuality risk). The operators of European Copenhagen - M1/M2 networks appear to be more exposed to Taipei - Wenhu Line this type of constraint. Paris - Line 14 Short distances between stations help Barcelona - Line 9 operators to maintain reasonable travel Dubai- Green line times between stations at low speed. They also limit infrastructure wear and reduce Vancouver - Canada Line the resulting maintenance costs. Vancouver - Expo Line Vancouver - Millenium Line Dubai - Red line

0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Average = 967m

Europe North America Middle East Asia

42 Commercial speed, a key Average Commercial Speed (in km/h) characteristic in transport Vancouver - Expo Line INNOVIA Métro service, is a core performance Vancouver - (Bombardier) assessment item Millenium Line Dubai - Red line Kinki Sharyo VAL rolling stock is the most represented Taipei - Wenhu Line VAL (Siemens) in the study sample, accounting for 35% of Copenhagen - M1/M2 Ansaldo Breda lines. It is operated at low speeds on some Paris - Line 14 MP14 (Alstom) lines (such as Turin) and high speeds on EMU Rotem (Hyundai/ others (Lille and Taipei). In addition, some Vancouver - Canada Line Mitsubishi) Nuremberg U2 DT1 DT2DT3 DT3-F operators appear to have experience with (Siemens) and a preference for a single type of roll- Lille - Line 2 ing stock (Ansaldo Breda for ATM, for Lille - Line 1 example). Toulouse - Line A VAL (Siemens) End-to-end line times depend mainly on Toulouse - Line B station stop times. Commercial speed is Rennes - Line A thus inversely proportional to the density Kinki Sharyo of stations on a given line (see previous Dubai - Green line slide vs. graph opposite). Paris - Line 1 MP05 (Alstom) Lyon - Line D MPL85 (Alstom) By optimizing acceleration and braking Série 9000 (Alstom) times, driverless metro lines increase the Barcelona - Line 9 average commercial speed of “dense” Milan - Line 5 AnsaldoBreda lines. DT3 DT3-F (Siemens) Low speed Nuremberg U3

High commercial speed entails additional Turin - Line 1 VAL (Siemens) London - Docklands B2007 (Bombardier) operating constraints. Operators manag- Light Railway ing low-speed lines only may find itdif - Lausanne - M2 Be 8/8 TL (Alstom) ficult to effectively operatehigh-speed 0 5 10 15 20 25 30 35 40 45 lines. Average = 33.1km/h

ATM Keolis Serco TL BCRTC SNC Lav. TBM TRTC GTT RATP Tisséo VAG

43 driverless metro lines generally provide improved access for passengers

Low Average High Accessibility Accessibility Accessibility

10% of the lines surveyed have 3% of the lines analyzed have 87% of the metro lines limited accessibility: fewer than average accessibility: more than analyzed have high 80% of the stations are equipped 80% of the stations are equipped accessibility: more than 90% for people with reduced mobility for people with reduced mobility of the total number of stations are equipped for passengers with reduced mobility

Station accessibility is generally related to infrastructure and thus to the transport authority’s approach to the issue. However, the availability of the implemented solutions results from the performance of operators. Besides the fact that this availability remains difficult to measure, the metro lines analyzed have high accessibility as a whole, resulting from standardized construction compliant with new standards and good rolling stock access (reduced distance between platform and train).

44 The choice of infrastructure Estimated Fraud Rate (in %) access is crucial to fraud rates. Open systems* fostering an Average = 4 % London - Docklands enhanced customer experience Light Railway OPEN SYSTEM lead to poor results in this Nuremberg - U3 respect Nuremberg - U2 Platform access gates are decisive in the fight against fraud. Milan - Line 5 The two networks still without these gates 34 - 45 Copenhagen - M1/M2 (Lille and Rennes) have fraud rates of over 10%, compared with 7% for networks with Paris - Line 1 / Line 14 Estimated fraud rate (%) gates. However, the DLR line in London remarkably has the lowest fraud rate Lausanne - M2 despite lacking access gates. Toulouse - Line A / Line B The installation of access gates is often a joint decision by the operator and the Barcelona - Line 9 transport authority. Access gates appear to be much more effective at limiting Turin - Line 1 fraud than an increase in (often random) ticket inspections by the operator. Vancouver - Canada Line

The socio-economic characteristics of Vancouver - Expo Line / the cities and neighborhoods served by Millenium Line the metro line are also a key factor in this Lyon - Line D respect, with the fraud rate varying on the same line from one station to the next. Rennes - Line A OPEN SYSTEM Operators are powerless against this vari- able but it does negatively impact their Lille - Line 1 / Line 2 OPEN SYSTEM operations over the long term. 0 2 4 6 8 10 12 14 16

Europe North America

* A system is considered as “open” if access to the metro network is not controlled by gates.

45 Service reliability and quality performance

French driverless metro Grades for Reliability and Service lines are extremely reliable, Quality Performance* (out of 5) a quality reinforced by Grade Punctuality**Ageingnetwork expertise in the operation London - Docklands Light Railway 4.2 99.11 % / 30 yrs of ageing networks Lille - Line 2 3.8 99.80 % / 28 yrs Lille - Line 1 3.8 99.80 % / 34 yrs Reliability and service quality are Lyon - Line D 3.8 97.23 % / 26 yrs key factors in driverless metro Taipei - Wenhu Line 3.8 99.73 % / 21 yrs 35 - 46 line operational performance, Rennes - Line A 3.8 99.66 % / 15 yrs determining: Toulouse - Line A 3.7 99.50 % / 24 yrs Paris - Line 14 3.5 99.70 % / 19 yrs // The passenger experience Dubai - Red line 3.3 99.23 % / 8 yrs // The rhythm of the service Dubai - Green line 3.3 99.83 % / 6 yrs Vancouver - Expo Line 3.0 96.00 % / 31 yrs Performance in this area is an oper- Toulouse - Line B 3.0 99.50 % / 10 yrs ational signature that also creates Copenhagen - M1/M2 3.0 98.80 % / 15 yrs the relational signature of the net- Turin - Line 1 3.0 99.90 % / 11 yrs work: a memorable mass transit Nuremberg - U3 2.8 99.00 % / 9 yrs Nuremberg - U2 99.00 % / 7 yrs experience or, on the contrary, an 2.8 Paris - Line 1 2.8 99.70 % / 6 yrs unpleasant experience encouraging Lausanne - M2 2.7 93.00 % / 9 yrs customers to use other transport Barcelona - Line 9 2.5 94.72 % / 8 yrs modes. Vancouver - Canada Line 2.2 96.00 % / 8 yrs Vancouver - Millenium Line 2.0 96.00 % / 15 yrs Milan - Line 5 1.8 96.00 % / 4 yrs 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

ATM Keolis Serco TL BCRTC SNC Lav. TBM TRTC GTT RATP Tisséo VAG

*The sum of the grade attributed to reliability and service quality (weight=1) and the grade attributed to the age of the infrastructure (weight=0.5) (5 corresponds to very old infrastructure). ** According to data availability and/or reliability, the figure may concern punctuality, availability or regularity (see glossary for definitions).

46 47 The age of the network and rolling stock has a strong impact on operational reliability

5 40 20 4.5 35 4 13 30 3.5 25 3 Average 3.4 20 2.5 15 2 10 1.5 5 Infrastructure and rolling stock age (in years) 1 0 Reliability and service quality performance (out of 5)

Lille - Line 1 Lille - Line 2 Paris - Line 1 Lyon - Line D Turin - Line 1 Milan - Line 5 Paris - Line 14 Nuremberg U2 Nuremberg U3 Lausanne - M2 Dubai - Red line Rennes - Line A Light Railway Toulouse - Line B Dubai - Green line Toulouse - Line A Barcelone - Line 9 London - Docklands Taipei - Wenhu Line Copenhague - M1/M2 Vancouver - Expo Line Vancouver - Canada Line Vancouver - Millenium Line

Performance infrastructure & Rolling stock Age of infrastructure

Vancouver - Millenium Line Infrastructure age is key to the reliability of For example, the ageing of rolling stock service. The deterioration of service qual- and stations has a direct impact on how ity is correlated to the ageing of assets passengers perceive cleanliness. (see graph above).

48 5 40 20 4.5 35 4 13 30 3.5 25 3 Average 3.4 20 2.5 15 2 10 1.5 5 Infrastructure and rolling stock age (in years) 1 0 Reliability and service quality performance (out of 5)

Performance infrastructure & Rolling stock Age of infrastructure

Vancouver - Millenium Line As it ages, infrastructure becomes less To ensure short headways, ageing infra- robust and limits the commercial speed structure and rolling stock require sub- of the line. stantial experience on the part of the operator. Failing which, maintenance costs rise drastically.

49 Highly strong variations in customer authority standards favoring the performance of certain lines and eroding that of others

Punctual operators are able to deliver punctuality PUNCTUAL performance only when headway requirements are longer. Any pressure to shorten those headways may considerably worsen their performance.

Paris 14 - Toulouse A - RATP Tisséo Dubai Red - Serco Lille 1 - Keolis 5 Dubai Green - Serco Turin - GTT Paris 1 - RATP Lille 2 - Keolis Punctual London DLR - Keolis Nuremberg - VAG Taipei - Wenhu Rennes - Line - TRT Toulouse B - Keolis Tisséo 4 Average Copenhagen - ATM Lyon D - Keolis

3 Milan - ATM

Vancouver - Canada Line - PBC Vancouver - Millenium Average 3.3 Vancouver - Expo Line - BCR Line - BCR

2 Barcelona - TBM Performance punctuality* (out of 5) Under-performers Lausanne M2 - TPL 1 0 1 2 3 4 5 Performance cadence* (out of 5) Performance cadence* (out of 5)

“Under-performers” are unable to ensure decent UNDER-PERFORMERS punctuality, even for long headways. Any transport- authority demands to shorten headways will significantly impact service levels.

Vancouver - Millenium Line

*Assessment method: the assessment is made comparatively between the lines analyzed. The maximum grade of 5 corresponds to the best headway/punctuality identified, the minimum being 1. Distribution within this range of grades is made in a proportional manner.

50 European operators, and French operators in particular, SUPER-PERFORMERS lead the way in terms of punctuality performance. Their experience, built on the margins of the high-level requirements of their respective transport authorities, brings them a major advantage over other operators.

Paris 14 - Toulouse A - RATP Tisséo Dubai Red - Serco Lille 1 - Keolis 5 Dubai Green - Serco Turin - GTT Paris 1 - RATP Lille 2 - Keolis London DLR - Keolis Nuremberg - VAG Taipei - Wenhu Rennes - Super-performers Line - TRT Toulouse B - Keolis Tisséo 4 Average Copenhagen - ATM Lyon D - Keolis

3 Milan - ATM

Vancouver - Canada Line - PBC Vancouver - Millenium Average 3.3 Vancouver - Expo Line - BCR Line - BCR

2 Barcelona - TBM Performance punctuality* (out of 5) Lausanne M2 - TPL Trying to perform 1 0 1 2 3 4 5 Performance cadence* (out of 5) Performance cadence* (out of 5)

TRYING TO PERFORM The short headways required by the transport authority appear to put the operators in the “trying to perform” category under pressure. BCRTC is currently struggling to deliver punctuality for short headways.

Vancouver - Millenium Line

While headways are generally the responsibility of transport authorities, their respect as part of a punctual service shows the ability of some operators to deliver high-level service under strong constraints. This is the case with the “Punctual” and “Super-performing” categories.

51 Although difficult to analyze Cleanliness Satisfaction Rate owing to its subjective nature, (positive opinion in %) satisfaction with cleanliness Average = 86.9% plays a crucial role in the Lyon - Line D experience of passengers Barcelona - Line 9 The satisfaction rate of users relative to 38 - 52 the cleanliness of the network and rolling Rennes - Line A Cleanliness satisfaction rate (positive opinion)* stock varies. It is closely linked to the (often subjective) perception of users. Taipei - Wenhu Line

While the satisfaction rate used in this Lausanne - M2 study corresponds to people having reported a positive opinion, the com- Paris - Lines 1 & 14 parison cannot be exact. This is because Toulouse - Lines A & B customer surveys are not the same for all the lines, which can bias comparisons to London - Docklands Light Railway a certain extent. Vancouver - Canada Line Ageing infrastructure and rolling stock can play an important role in the perception Copenhagen - M1/M2 of users. But the performance of operators on cleanliness remains decisive, as shown Dubai - Green & Red line* by the mediocre satisfaction ratings of Turin - Line 1 certain lines (including Turin and Milan).

Nuremberg - U2 & U3

Lille - Lines 1 & 2

Vancouver - Millenium & Expo Line

Milan - Line 5 0 60 70 80 90 100

ATM Keolis Serco TL BCRTC SNC Lav. TBM TRTC GTT RATP Tisséo VAG

*Satisfaction rate estimated via observations and comparisons between metro lines.

52 The safety of a metro line Safety Satisfaction Rate reflects user perceptions of the (positive opinion in %) network and the socio-economic Average = 80% situation of the urban areas Copenhagen - M1/M2 served Dubai - Green & Red line* The safety of a metro line depends directly on the surrounding environment. Taipei - Wenhu Line The safety performance of a metro line is London - Docklands Safety satisfaction rate (positive opinion) directly related to that of the city in ques- Light Railway tion, and more specifically thedifferent Nuremberg - U2 & U3* areas of that city. 39 - 53

Lines in cities such as Dubai, Copenhagen Lausanne - M2 and Taipei benefit from the high levels of Vancouver - Canada Line safety existing outside the metro in the city itself. Rennes - Line A*

Safety quality is often judged by users in Turin - Line 1 terms of the entire network. The grades of metro lines such as lines 1 and 14 in Paris, Lille - Lines 1 & 2 reputed to be the city’s safest, are lowered Vancouver - Millenium by the poor perception of the network & Expo Line as a whole. Lyon - Line D* Some lines serving areas far from the center with sensitive socio-economic sit- Paris - Lines 1 & 14 uations appear to be more exposed to a Barcelona - Line 9 drop in the overall safety level and, hence, a drop in user satisfaction (for example, Toulouse - Lines A & B Toulouse, Lyon and Lille). Milan - Line 5 0 60 70 80 90 100

ATM Keolis Serco TL BCRTC SNC Lav. TBM TRTC GTT RATP Tisséo VAG

*Satisfaction rate estimated via observations and comparisons between metro lines.

53 Channels for sharing information traffic information to other places, includ- (on travel times and traffic) in ing station interiors and even throughout real time are widespread, serving the passenger experience. to ensure the quality of the It is important for operators to multiply passenger experience passenger information media as these last are vital to a successful passenger experience. All the networks studied shared passen- To that end, open data are an ideal way ger information on platforms. This remains of transmitting information (e.g. social a preferred dissemination channel for media, airport displays, chatbots, etc.). operators. Some operators are working on this issue However, operators are seeking to extend (see following slide). ON THE PLATFORM MOBILE STATION ON BORD WEBSITE Paris – Lines 14 & 1 Barcelona – Line 9 Lausanne – M2 Lyon – Line D Rennes – Line A Toulouse (2 lines) Lille (2 lines) Nuremberg U2 & U3 Milan Line 5 Dubai (2 lines) Taipei – Wenhu line Turin – Line 1 Copenhagen (2 lines) London – DLR Vancouver (3 lines)

Traffic information is a pillar of the passenger experience in urban environments. Providing this information in real time and ensuring its reliability are both key to gaining the loyalty of customers through an augmented transport experience. In addition, opening metro lines up to competition and the arrival of new transport modes will require operators to differentiate. Providing multi-channel passenger information in real time is a way of doing just that.

54 The digital transformation and innovation are central to the group strategy of some operators

Digital maturity and innovation The driverless metro line benchmark iden- tifiesfive trends in innovation: Paris (RATP) // The reliability and sharing of passenger information and traffic news. Barcelona (TBM) This data is decisive to the passenger experience.

Rennes (Keolis) // Bringing passengers a mobile app with increasingly extensive function- alities and using it as a sales channel. Dubai (Serco) // The implementation of a reliable connection network (with a clear shift to Lille (Keolis) 4G rather than WiFi given the interest shown by telecom operators in Taipei (TRTC) this area) covering all stations and the entire journey. Most of the time Toulouse (Tisséo) spent by the passenger being on the train.

Nuremberg (VAG) // Open data, bringing operators a unique opportunity to boost innovation by opening systems up to the Lyon (Keolis) outside world.

// Big data for “transport as a service” Milan (ATM) (TAAS). Operators that know more about their customers can bring Lausanne (TL) these last a new, multimodal and personalized passenger experience. Analysis of passenger flows in Turin (GTT) stations can be used to more finely adjust supply to demand. London (Keolis)

Copenhagen (ATM)

Vancouver (BCRTC) 1.0 2.0 / 2.0+ 3.0

55 56 A detailed benchmark of the driverless metro lines under review

57 Barcelona: LINE 9

CHARACTERISTICS evaluation*

Infrastructure • Average commercial speed (29km/h) and long distances between sta- and rolling stock tions (>1,000m) 1 • Long headways, with 3-min intervals at rush hour 3.4 • Accessibility: good for PWRM (>80% of stops accessible) • Fraud rate: <4%

Service reliability and quality • Regularity: 94.72% (this is the contractual objective) • Cleanliness: 9.9 / 10 satisfaction rate for trains and stations 2 • Passenger information: information in real time disseminated in sta- 3.25 tions, scheduled times and traffic info on mobile app, on board and the TMB website

Auxiliary • Route planner app with real-time information (TMB app) • 4G coverage for entire line 9 and innovative • The operator, TMB, has rolled out an open-data platform with 6 inter- services faces. In this respect it remains less mature than operators such as 3 RATP and Keolis • TMB already uses big data to model passenger flows in stations and adapts line operation in consequence and in a predictive manner • In 2015 TMB introduced NFC and mobile payments for the entire urban network

KEY POINTS

• Line 9 is the first driverless line in Spain. It comprises 2 sections (L9 North and L9 South). Once these two sections have been joined up, it will be the longest driverless metro line in Europe.

*Raw grades out of 5, not weighted for age of network.

58 Age of network 8 years

City, Country Barcelona, Spain

Line Line 9

Wheel type Steel

Degree of automation GoA4

Operator Transports Metropolitans de Barcelona

Transport authority Metropolitan Transport Authority

Passengers per hour and per direction (PPHPD) 38,500

Signaling TrainGuard MT CBTC - Airlink

Network length in km 30

Number of stations in operation 24

Date commissioned 2009

Passengers per car 895 (passengers per train)

Manufacturer / model Alstom Série 9000

Average commercial speed 29km/h

*Age of network in number of years of driverless line operation.

59 COPENHAGEN: LINES 1 AND 2

CHARACTERISTICS evaluation*

Infrastructure • High commercial speed (40km/h) and average distances between sta- and rolling stock tions (>1,000m) 1 • Average headways, with 2-min intervals at best at rush hour 3.8 • Accessibility: good for PWRM (>80% of stops) • Fraud rate: 2.7%

Service reliability • Respect of departure numbers: 98.7% • Cleanliness: more than 80% of users say they are satisfied or very sat- and quality isfied with the cleanliness of the stations and trains 2 • Passenger information: real-time information on platforms and 3.5 planned hours on the Metro Services website (80% of people say they are satisfied or very satisfied with information updates) • Passenger safety: more than 95% of users say they feel safe

Auxiliary and innovative services • 3G available on entire metro line, including the 22 stations 3 • The Copenhagen network doesn’t have a dedicated mobile app

KEY POINTS

• The operator, Metro Service, is a joint venture between ATM and Ansaldo STS. • The manufacturer is thus likely to provide ATM with strong support in rolling stock maintenance. The current operating and maintenance contract terminates on December 31, 2018 and the new contract covers the 2019-2024 period. • The Copenhagen metro won the “World’s Best Driverless Metro” award in 2008, 2009 and 2010. • The Copenhagen metro runs 24-7.

*Raw grades out of 5, not weighted for age of network.

60 Age of network* 15 years

City, Country Copenhagen, Danemark

Line M 1 M 2

Wheel type Steel

Degree of automation GoA4

Operator Metro Service

Transport authority Movia

Passengers per hour and per direction (PPHPD) 12,000

Signaling Ansaldo STS driverless solution - CBTC

Network length in km 21.5

Number of stations in operation 22

Date commissioned 2002

Passengers per car 200 - 34 trains, 3 cars per train

Manufacturer / model Ansado Breda

Average commercial speed 40km/h

*Age of network in number of years of driverless line operation.

61 DUBAi: Line RED

CHARACTERISTICS evaluation*

Infrastructure • High commercial speed (42km/h) but long distances between stations and rolling stock (>1,800m) • Long headways, with 3-min intervals at rush hour 1 • Accessibility: 100% for PWRM 3.6 • Fraud rate: unknown • Free park-and-ride at three stations

Service reliability and quality • Strong punctuality (>99%) • RTA does not disclose passenger satisfaction levels regarding cleanli- 2 ness, but numerous opinions point to high levels of cleanliness 4.5 • Passenger information: real-time information disseminated in stations, on platforms, on board; scheduled times on the RTA website

Auxiliary and innovative • Network fully covered by mobile phone network services • WiFi access across entire network 3 • NFC payments by mobile phone available since 2013 • The Dubai metro stands out through its high levels of in-train connectivity: Very high quality WiFi and 4G connectivity end-to-end

KEY POINTS

• Serco has obtained the renewal of the operating and maintenance contract (terminating in 2019) for two additional years • In addition to offering a more competitive offer, SERCO has committed to making a skills transfer, in particular by recruiting more nationals in operations, maintenance and supervision

*Raw grades out of 5, not weighted for age of network.

62 Age of network* 8 years

City, Country Dubai, UAE

Line Red Line

Wheel type Steel

Degree of automation GoA4

Operator Serco

Transport authority RTA

Passengers per hour and per direction (PPHPD) 26,000

Signaling SelTrac CBTC - UTO

Network length in km 52.1

Number of stations in operation 29

Date commissioned 2009

Passengers per car 110 - 62 five-car trains (550 pass cap)

Manufacturer / model Kinki Sharyo

Average commercial speed 42km/h

*Age of network in number of years of driverless line operation.

63 Dubai: Line GREEN

CHARACTERISTICS evaluation*

Infrastructure • Average commercial speed (32km/h) and long distances between sta- and rolling stock tions (>1,100m) 1 • Very long headways, with 6-min intervals at rush hour 3.4 • Accessibility: 100% for PWRM • Fraud rate: unknown, estimated on the basis of observations

KEY POINTS

• Serco has obtained the renewal of the operating and maintenance contract (terminating in 2019) for two additional years. • In addition to offering a more competitive offer, SERCO has committed to making a skills transfer, in particular by recruiting more nationals in operations, maintenance and supervision.

*Raw grades out of 5, not weighted for age of network.

64 Age of network* 6 years

City, Country Dubai, UAE

Line Green Line

Wheel type Steel

Degree of automation GoA4

Operator Serco

Transport authority RTA

Passengers per hour and per direction (PPHPD) 26,000

Signaling SelTrac CBTC - UTO

Network length in km 22.5

Number of stations in operation 20

Date commissioned 2011

Passengers per car 110 - 25 five-car trains (550 pass cap)

Manufacturer / model Kinki Sharyo

Average commercial speed 32km/h

*Age of network in number of years of driverless line operation.

65 LAUSANNE: Line M2

CHARACTERISTICS evaluation*

Infrastructure • Low commercial speed (18km/h) – major gradients with an incline as and rolling stock steep at 12% – but short average distance between stations (<500m) • Average headways, with 2.5-min intervals at rush hour 1 • Accessibility: high for PWRM 3.4 • Fraud rate: ~ 3% • One park-and-ride facility

Service reliability • The operator has no quantified data on punctuality but does disclose a user satisfaction rate, of 93/100 and quality • Cleanliness: user satisfaction rate of 95/100 2 • Passenger information: real-time information in stations, on platforms, 3.5 via mobile apps, and scheduled times/traffic info on the TL website (satisfaction rate with information: 94%) • Passenger safety: satisfaction rate of 84%

Auxiliary and innovative services • Route planner app with real-time information (TL Live app) 3 • TL has chosen SMS payments rather than NFC technology

KEY POINTS

• Line M2 is the first and only driverless metro line in Switzerland, opened in 2008. In 2012 Métro Lausanne-Ouchy SA, concession-holder and long-standing operator, was struck off the Commercial Register and its assets taken over by Transports Publics Lausannois. This change in operator was made following congestion problems resulting from an under-estimation of user numbers in preparatory studies. • A project has been initiated to increase capacity by upgrading the line’s automatic systems.

*Raw grades out of 5, not weighted for age of network.

66 Age of network* 9 years

City, Country Lausanne, Switzerland

Line M2

Wheel type Pneumatic

Degree of automation GoA4

Operator Transports publics lausannois

Transport authority -

Passengers per hour and per direction (PPHPD) 5,600

Signaling Urbalis 300 CBTC

Network length in km 6

Number of stations in operation 14

Date commissioned 2008

Passengers per car 222 - 15 x 2-car trains

Manufacturer / model Alstom

Average commercial speed 18km/h

*Age of network in number of years of driverless line operation.

67 Lille: LineS 1 and 2

CHARACTERISTICS evaluation*

Infrastructure • High commercial speed (35km/h) given the age of the rolling stock. Short average distance between stations (750m) and rolling stock • The shortest rush-hour headways in the world: ~ 66 seconds • High fraud rate (~14%) owing to open network. This is expected to fall 3.8 1 with the installation of access gates throughout the network, to be completed by 2020. • Accessibility: 100% for PWRM

Service reliability • Availability: 99.8% / Respect of journey time: 99% • Cleanliness: satisfaction rate of 78% for stops and 82% for trains and quality • Passenger information: real-time information in stations, on platforms, via mobile apps, and scheduled times/traffic info on the Transpole 2 website. Satisfaction rate of 86% for availability of information and 3.25 66% for information on line disruptions • Passenger services: stores, Internet connection to public services • Passenger safety: satisfaction rate of 73% for stops and 80% for trains

Auxiliary • Route planner app with real-time information (Transpole app) • Launch of Moodi, an app designed to generate real-time mapping of and innovative the “emotional climate” of passengers (passengers can express how services they feel and report on incivility or violence) • 3G/4G will be available in 2018 throughout both lines 3 • The operator favors NFC technology, providing customers with an app that can recharge travel passes via mobile phone. This is a further step towards 100% electronic travel • Passengers have access to a “PlanBookTicket” app as part of a multimodal transport approach

KEY POINTS

• Lille was the world’s first urban driverless metro line, commissioned in 1983. • Transpole has received excellent feedback on its rolling stock, this last serving to optimize maintenance processes. • Maintenance costs are falling steadily, while the Mean Kilometer Between Failure (MKBF) rate has risen constantly in the last few years (currently at 5,600km).

*Raw grades out of 5, not weighted for age of network.

68 Line 1: 34 years Age of network* Line 2: 28 years

City, Country Lille, France

Line Line 1 Line 2

Wheel type Pneumatic

Degree of automation GoA4

Operator Transpole

Transport authority Métropole Européenne Lilloise

Passengers per hour and per direction (PPHPD) 11,000

Signaling TrainGuard MT

Network length in km 12.6 31.5

Number of stations in operation 18 44

Date commissioned 1983 1989

Passengers per car 156

Manufacturer / model Alstom/Siemens - VAL 208 / VAL 206

Average commercial speed 34km/h

*Age of network in number of years of driverless line operation.

69 London: DOCKLANDS LIGHT RAILWAY

CHARACTERISTICS evaluation*

Infrastructure • Short average distance between stations (750m) and rolling stock • Long headways, with 5-min intervals at rush hour 1 • Accessibility: good for PWRM (>80% of stops) 3.2 • Fraud rate: 0.6%, down sharply since Keolis Amey took over the franchise

Service reliability • Punctuality: 99.11% and quality • Cleanliness: the composite cleanliness indicator is 88/100 2 • Passenger information: real-time information on platforms, and sched- 3.75 uled times/traffic info on the TFL website • Passenger safety: 89%

Auxiliary and innovative • Monthly reporting available on the performance of the network, dis- services seminated by TFL 3 • Traffic information and travel times available in real time via the TFL transport authority

KEY POINTS

• The concession was recently taken over (2014) by the Amey group and the French urban transport player Keolis. Consequently, the performance indicators for the line could soon change in the future.

*Raw grades out of 5, not weighted for age of network.

70 Age of network* 30 years

City, Country London, UK

Line Docklands Light Railway

Wheel type Steel

Degree of automation GoA3

Operator KeolisAmey Docklands

Transport authority Transport for London

Passengers per hour and per direction (PPHPD) 6,000

Signaling SelTrac CBTC (Thales)

Network length in km 34

Number of stations in operation 45

Date commissioned 1987

Passengers per car 284

Manufacturer / model Bombardier

Average commercial speed 18km/h

*Age of network in number of years of driverless line operation.

71 Lyon: Line D

CHARACTERISTICS evaluation*

Infrastructure • Average commercial speed (30km/h) and relatively short average dis- and rolling stock tance between stations (890m) 1 • Average headways, with 2-min intervals at rush hour 3.6 • High accessibility rate for PWRM (superior to contractual objectives) • Fraud rate: ~6.5% (closed network)

Service reliability • Regularity: 97.23% (superior to contractual objective) • Cleanliness: Metro stations: 99.54% / Metro cars: 99.20% (superior to and quality contractual objective) • Passenger information: information in real time and disruptions com- 2 municated in stations, on platforms, and on mobile phones; scheduled 3.75 times and traffic info on the Sytral website (Metro stations: 99.54% / Metro cars: 99.20% (superior to contractual objective) • Passenger safety: estimated at 75%

Services auxiliaires et innovants • Route planner app with real-time information (TCL app) 3 • Implementation of WiFi by end-2017 in stations (but not on entire line) • More than 15 APIs available via the Greater Lyon open-data platform

KEY POINTS

• Line D was the world’s first driverless driverless metro line with train intervals managed by a CBTC system. The operator is planning to increase rush-hour capacity by 15% by 2020 (to this end, in 2015 Sytral and Keolis Lyon tested the operation of a train with 4 cars as opposed to the current 2). • The principle of full automation for metro line B in 2019 has been accepted by Sytral.

*Raw grades out of 5, not weighted for age of network.

72 Age of network* 26 years

City, Country Lyon, France

Line Line D

Wheel type Pneumatic

Degree of automation GoA4

Operator Keolis Lyon

Transport authority SYSTRAL

Passengers per hour and per direction (PPHPD) 15,000

Signaling TrainGuard MT CBTC

Network length in km 12.5

Number of stations in operation 15

Date commissioned 1991

Passengers per car 291

Manufacturer / model Alstom MPL 85

Average commercial speed 30km/h

*Age of network in number of years of driverless line operation.

73 MILAN: Line 5

CHARACTERISTICS evaluation*

Infrastructure • Modest commercial speed (30km/h) but short distances between sta- and rolling stock tions (>1,000m) • Relatively long headways, with 3-min intervals at rush hour 1 • Accessibility: good for PWRM (>80% of stops) 4 • Low fraud rate (~1%) due to ticket access gates at network entrances and exits

Service reliability • Strong punctuality (>99%) • No contractual commitment to cleanliness in delegation contract. and quality Cleanliness levels are measured by a satisfaction survey that must respect a minimum threshold. The latest known and available rates are 2 particularly low. 2.25 • Passenger information: real-time information on platforms, in-train, via mobile apps, and scheduled times/traffic info on the ATM website. • Passenger safety: 61% for latest known and available rates

Auxiliary and innovative services • Route planner app with real-time information (ATM app) 3 • 3G is available on all the metro lines • Electronic travel pass via mobile app (non-NFC)

KEY POINTS

• Line 5 is subject to a concession contract between the City of Milan and the Metro 5 SpA company, whereby the City concedes the financing, design, production and operation/maintenance to Metro 5 SpA. For the account of Metro 5 SpA, ATM operates and maintains Line 5 via a gross-cost contract applying for the entire duration of the concession through end-2040. • ATM is responsible for all regular maintenance of rolling stock (Ansaldo’s activity is limited to exceptional work as provided for in the warranty).

*Raw grades out of 5, not weighted for age of network.

74 Age of network* 4 years

City, Country Milan, Italy

Line Line 5

Wheel type Steel

Degree of automation GoA4

Operator Azienda Transporti Milanesi

Transport authority Azienda Transporti Milanesi

Passengers per hour and per direction (PPHPD) 10,700

Signaling Ansaldo STS driverless solution - CBTC

Network length in km 12.6

Number of stations in operation 19

Date commissioned 2013

Passengers per car 134 - 21 trains, 4 cars per train (536 passengers per train)

Manufacturer / model Hitachi Rail Italy (formerly Ansado Breda) / Driverless Metro

Average commercial speed 30km/h

*Age of network in number of years of driverless line operation.

75 NUREMBERG: Line U2

CHARACTERISTICS evaluation*

Infrastructure • High commercial speed (34.4km/h) and relatively short average dis- and rolling stock tance between stations (880m) • Relatively long rush-hour headways: ~3-min 1 • Very low fraud rate of 0.9% (ranking in the top three networks in this 4 study) • Accessibility: 100% for PWRM

Service reliability • High punctuality rate of 99% and quality • Cleanliness: satisfaction rate of 80% for stations and rolling stock • Passenger information: widely available. VAG even provides an inter- 2 face providing the real-time geolocation of the trains 3.75 • Passenger safety: passengers are generally satisfied with the safety level of the Nuremberg network, though there are some incidents

Auxiliary and innovative • Route planner app with real-time information services • VAG is working with T-System on a big data project to analyze passen- 3 ger flows across the entire transport network • WiFi is available in stations and on trains

KEY POINTS

• Line U2 has high-level punctuality and a very low fraud rate, characteristic of German networks.

*Raw grades out of 5, not weighted for age of network.

76 Age of network* 7 years

City, Country Nuremberg, Germany

Line Line U2

Wheel type Steel

Degree of automation GoA4

Operator VGN / VAG

Transport authority Nuremberg Public Transport

Passengers per hour and per direction (PPHPD) 12,800

Signaling TrainGuard MT

Network length in km 13.2

Number of stations in operation 16

Date commissioned 2010

Passengers per car 128

Manufacturer / model Siemens Munich A

Average commercial speed 34.4km/h

*Age of network in number of years of driverless line operation.

77 NUREMBERG: Line U3

CHARACTERISTICS evaluation*

Infrastructure • Low commercial speed (26km/h) and relatively short average distance and rolling stock between stations (650m) • Relatively long rush-hour headways: ~3-min 1 • Very low fraud rate of 0.9% (ranking in the top three networks in this 3.8 study) • Accessibility: 100% for PWRM

KEY POINTS

• Line U3 has high-level punctuality and a very low fraud rate, characteristic of German networks.

*Raw grades out of 5, not weighted for age of network.

78 Age of network* 9 years

City, Country Nuremberg, Germany

Line Line U3

Wheel type Steel

Degree of automation GoA4

Operator VGN / VAG

Transport authority Nuremberg Transport Public

Passengers per hour and per direction (PPHPD) 12,800

Signaling TrainGuard MT

Network length in km 6.5

Number of stations in operation 11

Date commissioned 2008

Passengers per car 128

Manufacturer / model Siemens Munich A

Average commercial speed 26km/h

*Age of network in number of years of driverless line operation.

79 pARIS: Line 1

CHARACTERISTICS evaluation*

Infrastructure • Average commercial speed (30km/h) and short distance between sta- and rolling stock tions (660m) 1 • Short maximum headways, with 2-min intervals at rush hour 3.4 • Accessibility: low for PWRM (<70% of stops) • Fraud rate: 3% for entire network (closed network)

Service reliability • Availability: >100% (% of real number of metro trains in circulation in rush hour relative to service ordered) and quality • RATP does not disclose information on passenger satisfaction concern- ing cleanliness but STIF reports that 89.6% of the stations comply with 2 the reference service level 3.75 • Passenger information: real-time information in stations, on platforms, in-train, via mobile apps, and scheduled times/traffic info on the RATP website

Auxiliary • Route planner app with real-time information (RATP app) and innovative • Most of the stations on Line 1 have 3G and 4G coverage services • An API console with more than 20 interfaces 3 • Big data and IoT are strategic topics for RATP and central to the group’s overall strategy • RATP is carrying out tests on the simulation and projection of passenger flows in stations, using big data

KEY POINTS

• Line 1 was one of the world’s first metro lines to be automated, in 2011. • Line 1 demonstrated its adaptability in summer 2015, absorbing the increase in passengers when the RER A line was shut down for work. To achieve headways of 100 seconds, Line 1 benefited from the temporary use of 4 shuttles intended for use on Line 14.

*Raw grades out of 5, not weighted for age of network.

80 Age of network* 6 years

City, Country Paris, France

Line Line 1

Wheel type Pneumatic

Degree of automation GoA4

Operator RATP

Transport authority STIF

Passengers per hour and per direction (PPHPD) 25,000

Signaling TrainGuard MT CBTC

Network length in km 16.5

Number of stations in operation 25

Date commissioned 2011

Passengers per car 120 (6 cars per train)

Manufacturer / model Alstom

Average commercial speed 30km/h

*Age of network in number of years of driverless line operation.

81 pARIS: Line 14

CHARACTERISTICS evaluation*

Infrastructure • High commercial speed (40km/h) and average distance between sta- and rolling stock tions (1,000m) 1 • Short maximum headways, with 85-sec intervals at rush hour 4.4 • Accessibility: good for PWRM (>80% of stops) • Fraud rate: 3% for entire network (closed network)

Service reliability • Availability: >100% (% of real number of metro trains in circulation at rush hour relative to the service ordered) and quality • RATP does not disclose information on passenger satisfaction concern- ing cleanliness but STIF reports that 89.6% of the stations comply with 2 the reference service level 3.75 • Passenger information: real-time information in stations, on platforms, in-train, via mobile apps, and scheduled times/traffic info on the RATP website

KEY POINTS

• The first large-size 100% driverless line in the world when it went into service, Line 14 is to be extended to link up the center of the capital with the Saint- Denis Pleyel business hub to the north and Orly airport to the south. • These extensions will be made while maintaining the current characteristics of the line, with a speed of over 40km/h.

*Raw grades out of 5, not weighted for age of network.

82 Age of network* 19 years

City, Country Paris, France

Line Line 14

Wheel type Pneumatic

Degree of automation GoA4

Operator RATP

Transport authority STIF

Passengers per hour and per direction (PPHPD) 25,000

Signaling TrainGuard MT CBTC

Network length in km 9.2

Number of stations in operation 9

Date commissioned 1998

Passengers per car 720 (passengers per train)

Manufacturer / model Alstom

Average commercial speed 40km/h

*Age of network in number of years of driverless line operation.

83 RENNES: Line A

CHARACTERISTICS evaluation*

Infrastructure • Average commercial speed (32km/h) and short distance between sta- and rolling stock tions (610m) 1 • Short headways, with 100 sec intervals at rush hour 4 • Accessibility: 100% for PWRM • High fraud rate (~10.4%) due to open network, to be closed in 2019/2020

Service reliability • Punctuality: 99.66%, system availability rate: 99.72% and quality • Cleanliness: satisfaction rate of 94.4/100 • Passenger information: real-time information in stations, on platforms, 2 via mobile apps, and scheduled times/traffic info on the Star website. 4.25 Satisfaction rate of 95/100 relative to the condition and the availability of information • Passenger safety: estimated at 80%

Services auxiliaires et • The 4G system of mobile telephony operators will be available from June 2017 on the entire Line A innovants • A customer mobility solution (mobile app) is available. It brings users a 3 route planning resource and real-time traffic information • In collaboration with OpenDataSoft, Keolis provides a comprehensive open-date platform with over 30 APIs for third-party developers

KEY POINTS

• The construction of a second driverless metro line (using the CityVal variant of Neoval) is under way, with service planned to begin in 2020. • The current delegation contract expires at end-2017 and will need to be reviewed before the inception of the second metro line, which will significantly change the Rennes transport offering.

*Raw grades out of 5, not weighted for age of network.

84 Age of network* 15 years

City, Country Rennes, France

Line Line A

Wheel type Pneumatic

Degree of automation GoA4

Operator Keolis Rennes

Transport authority Rennes Métropole

Passengers per hour and per direction (PPHPD) 6,000

Signaling TrainGuard MT CBTC (Siemens)

Network length in km 8.56

Number of stations in operation 15

Date commissioned 2002

Passengers per car 156

Manufacturer / model Siemens

Average commercial speed 32km/h

*Age of network in number of years of driverless line operation.

85 Taipei: WENHU Line (BROWN LINE)

CHARACTERISTICS evaluation*

Infrastructure • High commercial speed (42km/h) and long distances between stations and rolling stock (>1.100m) 1 • Maximum headway of 2-min at rush hour 4.2 • High level of accessibility for PWRM • Fraud rate: unknown

Service reliability and quality • Punctuality: 99.73% • Cleanliness: satisfaction rate of over 95/100 2 • Passenger information: real-time information on platforms and in 4.25 trains, and 24-7 customer hotline

Auxiliary • WiFi accessible in all stations and innovative • NFC contactless technology was rolled out across the network starting services in 2012 3 • The Taipei metro has an open-data platform but does not communicate real-time data. The interface is mainly used to share monthly statistics • MRT provides a mobile app with offline information, with no route- planner function

KEY POINTS

• The Taipei metro is considered by several local press sources to rank among the top 4 most efficient networks in the world (with Hong Kong, Seoul and Singapore). This is a result of top-level punctuality, very good condition infrastructure and high-level cleanliness.

*Raw grades out of 5, not weighted for age of network.

86 Age of network* 21 years

City, Country Taipei, Taiwan

Line Wenhu line

Wheel type Pneumatic

Degree of automation GoA4

Operator Taipei Rapid Transit

Transport authority -

Passengers per hour and per direction (PPHPD) 28,400

Signaling Bombardier CityFlo

Network length in km 25.2

Number of stations in operation 24

Date commissioned 1996

Passengers per car 114

Manufacturer / model Alstom-Bombardier / VAL

Average commercial speed 42km/h

*Age of network in number of years of driverless line operation.

87 TOULOUSE: LineS A and B

CHARACTERISTICS evaluation*

Infrastructure • Average commercial speed (33km/h) but relatively short distance 4.4 and rolling stock between stations (740m) Line A 1 • Short headways, with 80 sec intervals at rush hour • Low fraud rate of ~3.2% (closed network) • Accessibility: good for PWRM (over 80% of stops accessible) 4.2 Line B

Service reliability • The operator does not share quantified data on punctuality, cleanliness or passenger information and quality • But these indicators are compliant with the quality commitment established with AFNOR, which has awarded Tisséo with NF Service 2 certification 3.5 • Passenger information: real-time information in stations, and scheduled times/traffic info on the Tisséo website • Passenger safety: Estimated at 67%

Services auxiliaires et • Route planner app with real-time information (Tisséo app) innovants • The 4G system of mobile telephony operators will be available from 3 October 2017 throughout lines A and B • An open-data platform is available. But the number of APIs remains low (10 interfaces)

KEY POINTS

• The line is operated by Tisséo-Réseau Urbain, a public undertaking of an industrial and commercial nature (EPIC). • Given the demographic and economic growth of the Toulouse agglomeration and its mobility needs, the plan is to double the capacity of line A between now and 2019.

*Raw grades out of 5, not weighted for age of network.

88 Line A: 24 years Age of network* Line B: 10 years

City, Country Toulouse, France

Line Line A Line B

Wheel type Pneumatic

Degree of automation GoA4

Operator Tisséo

Transport authority Tisséo-EPIC

Passengers per hour and per direction (PPHPD) 7,000

Signaling TrainGuard MT CBTC

Network length in km 12.5 15

Number of stations in operation 18 20

Date commissioned 1993 2007

Passengers per car 200

Manufacturer / model Siemens (Val 206 and Val 208)

Average commercial speed 34km/h 33km/h

*Age of network in number of years of driverless line operation.

89 Turin: Line 1

CHARACTERISTICS evaluation*

Infrastructure • Low commercial speed (25.5km/h) but relatively short distance and rolling stock between stations (660m on average) 1 • Long average headways, with 3-min intervals at rush hour 3.8 • Very high level of accessibility for PWRM • Fraud rate of 4%, moderately high for a closed network

Service reliability and quality • Service availability is very high, close to 99.9% 2 • Passenger information: available on platforms and in-train 3.5

Auxiliary and innovative services • A mobile app is available. It has route-planner functions but no real- 3 time traffic information

KEY POINTS

• The Turin metro network is recent and the infrastructure is new and well maintained. This explains in part the reliability of the service and the strong satisfaction of customers regarding cleanliness. • The fraud rate is considered as moderately high for a closed network. This rate does not fall within the average in Italy, which appears to vary between 1% and 3%.

*Raw grades out of 5, not weighted for age of network.

90 Age of network* 11 years

City, Country Turin, Italy

Line 1

Wheel type Pneumatic

Degree of automation GoA4

Operator Gruppo Trasporti Torinesi

Transport authority -

Passengers per hour and per direction (PPHPD) 23,000

Signaling TrainGuard MT CBTC

Network length in km 13.2

Number of stations in operation 21

Date commissioned 2006

Passengers per car 200

Manufacturer / model Siemens / VAL 208

Average commercial speed 25.5km/h

*Age of network in number of years of driverless line operation.

91 VANCOUVER: CANADA Line

CHARACTERISTICS evaluation*

Infrastructure • High commercial speed (35km/h) but long average distance between and rolling stock stations (1,300m) 1 • Long headways, with 6-min intervals at rush hour 3.2 • Accessibility: good for PWRM (>80% of stops) • Fraud rate: 5.4%

Service reliability • The operator does not disclose figures on punctuality, but this last is judged as satisfactory or very satisfactory by 92% of passengers and quality • Cleanliness: 87% of passengers give a good to excellent rating of cleanliness in trains and in stations 2 • Passenger information: real-time information disseminated on plat- 2.75 forms; scheduled times on the Translink website • Passenger safety: 82% of passengers give a good to excellent rating of on-board safety

Auxiliary and innovative services 3 • No dedicated app, but a route planner developed for mobiles

KEY POINTS

• The Canada Line was built as part of a public-private partnership. SNC Lavalin is the operator for a 35-year period, with minimum frequency guaranteed by Translink.

*Raw grades out of 5, not weighted for age of network.

92 Age of network* 8 years

City, Country Vancouver

Line Canada Line

Wheel type Steel

Degree of automation GoA4

Operator Protrans BC

Transport authority Greater Vancouver Transport Authority

Passengers per hour and per direction (PPHPD) 15,000

Signaling SelTrac CBTC - UTO

Network length in km 19.2

Number of stations in operation 16

Date commissioned 2009

Passengers per car 334

Manufacturer / model Hyundai Rotem

Average commercial speed 35km/h

*Age of network in number of years of driverless line operation.

93 VANCOUVER: EXPO Line / MILLENIUM Line

CHARACTERISTICS evaluation*

Infrastructure and rolling stock • High commercial speed (45km/h) but long distance between stations 4 (>1,500m) Expo Line 1 • Average headways, with 2-min intervals at rush hour • Fraud rate: 5.5% 3.6 Millenium Line

Service reliability • The operator does not disclose figures on punctuality, but this last is judged as satisfactory or very satisfactory by 81% of passengers and quality • Cleanliness: 71% of passengers give a good to excellent rating of cleanliness in trains and in stations 2 • Passenger information: real-time information disseminated on plat- 2 forms; scheduled times on the Translink website • Passenger safety: 76% of passengers give a good to excellent rating of on-board safety

Auxiliary and innovative services 3 • No dedicated app, but a route planner developed for mobiles

KEY POINTS

• British Columbia Rapid Transit Company Ltd (BCRTCTC) operates and maintains the two Skytrain lines on behalf of TransLink.

*Raw grades out of 5, not weighted for age of network.

94 Expo Line: 31 years Age of network* Millenium Line: 15 years

City, Country Vancouver, Canada

Line Skytrain Expo Line Skytrain Millenium Line

Wheel type Steel

Degree of automation GoA4

Operator British Colombia Rapid Transit Company

Transport authority Autorité des Transports du Grand Vancouver

Passengers per hour and per direction (PPHPD) 16,000 15,000

Signaling SelTrac CBTC - UTO

Network length in km 28.9 15

Number of stations in operation 20

Date commissioned 1985 2002

Passengers per car 500 (passengers per train) 200

Manufacturer / model Bombardier Transportation Siemens (Val 206 and Val 208)

Average commercial speed 44km/h 33km/h

*Age of network in number of years of driverless line operation.

95 APPENDICES

96 Indicator assessment methodology

Performance is graded in 5 levels so as to determine for each indicator the relative performance of each network in the benchmark

Performance level Scored out of 5 attribution of grade

Mediocre performance

The least efficient network in the scope 1

Grade attributions are Insufficient performance 2 systematically objectified on the basis of the quantitative data compiled Average performance 3 Where quantified data are not available, the grade is based on all the supplementary Good performance qualitative items liable to 4 enlighten the analysis

Excellent performance

The most efficient network in the scope 5

97 The comparative analysis of the metro lines is based on indicators taking account of infrastructure, service quality and the age of the assets

Performance of infrastructure Commercial speed and rolling stock Distance between stations Grades for the performanc Headway of infrastructure Accessibility and rolling stock Intrusion rate in network 1 0.5

The effect of age Overall performance Grades for the performance on the infrastructure of the metro line and rolling stock of driverless metro lines

Punctuality 1.5 Grades for reliability Cleanliness and service quality Passenger information performance Passenger safety Service reliability and quality

98 Performance of infrastructure Commercial speed and rolling stock Distance between stations Grades for the performanc Headway of infrastructure Accessibility and rolling stock Intrusion rate in network 1 0.5

The effect of age Overall performance Grades for the performance on the infrastructure of the metro line and rolling stock of driverless metro lines

Punctuality 1.5 Grades for reliability Cleanliness and service quality Passenger information performance Passenger safety Service reliability and quality

99 Communication-based train control (CBTC) architecture

The main characteristics of the CBTC system are: • the localization of trains independently of track circuits; • high-speed dual-direction transmission between ground equipment and trains; • a system comprising computers located both on ground and in trains.

Automatic Train Supervision (ATS) serves as an interface between ATS the operators and the system. It can be used to control and glob- Train supervision ally supervise the system and manage train flows in respect of theoretical service.

Wayside ATP: manages communications with the trains in its area and calculates the movement limits to be respected by the Wayside ATP + ATO trains systems Wayside ATO: controls the destination of trains and the parame- ters for regulating their movement (station stop times)

Several technologies exist for communications between onboard systems and ground systems, including induction loops, beacons and radio communication. This last technology is the most widely used

Onboard ATP: continuous control of train speed consistent with safety criteria (braking management where necessary). Serves to Onboard ATP + ATO communicate the data necessary to operations with the Wayside systems ATP Onboard ATO: automatic control of the acceleration and braking of the train, consistent with theoretical service

ATP: Automatic Train Protection ATO: Automatic Train Operation

100 Glossary

Automated guideway transit (AGT): a GoA 3 (Grade of automation 3): system similar to the metro, in that it runs driverless operation in which the on a dedicated circuit which does not train is started and stopped auto- interfere with other modes of transport. matically but an operator (not a It differs from a metro by often being com- driver) is responsible for controlling posed of one car per train, with a maxi- the doors and operating incidents mum capacity of 100 passengers. Through and is called on to drive the train its simple design, AGT does not have the where necessary same limitations as a metro. GoA 4 (Grade of automation 4): operation with no human resources Automatic train protection (ATP), auto- on board, in which the train is started matic train control (ATC), automatic train and stopped automatically and the operation (ATO): onboard systems serving control of the doors and operating respectively to prevent collisions, control incidents are fully automated. route setting and train regulation, and control train acceleration and deceleration. Availability: an indicator quantifying the difference between the actual number of Automation level: the level of automa- kilometers traveled and the number of tion is defined according to the division theoretical kilometers. Availability is cal- of responsibilities between the system and culated as the ratio between the actual human resources of the activities relating number of kilometers traveled and the to the operation of the service. number of theoretical kilometers. GoA 1 (Grade of automation 1): 100% manual operation in which the Communication-based train control driver is responsible for starting and (CBTC) system: an automatic rail transport stopping the train and controlling the control system (train or metro) based on doors and operating incidents. continuous communication between the GoA 2 (Grade of automation 2): train and the computers used to coordi- semi-automatic operation in which nate traffic. the train is started and stopped auto- Conventional metro (as defined for the matically but the driver is responsi- study): a metro (see definition above) ble for controlling the doors and whose running is entirely or partially operating incidents and is called on ensured by a driver (grade 1 and grade 2) to drive the train where necessary

101 Driverless metro (as defined for the People mover: a fully automated light study): a mass transit system operating transit rail system. The term is generally on a dedicated underground or over- used only for raised, single-rack shuttles ground circuit powered by rail and trans- operating over short distances at airports, porting passengers essentially making leisure parks or compact urban areas. back-and-forth journeys/commutes. the number of passengers per Generally speaking, the train is made up PPHPD: hour and per direction is a measurement of several cars (up to 6 or 8), comprises of the capacity of a public transport a raised access level, and has a capacity system. PPHPD is used to determine the of over 100 passengers. The train is oper- capacity requirements of the rolling stock ated entirely automatically without driver (and thus the CAPEX and OPEX of the input (grade 3 or grade 4). The LTR, AGT project). and people-mover systems defined in this glossary do not fall within this category. Punctuality: an indicator quantifying the difference between the actual time of Driverless metro operators: TL (Transports arrival at a station and the theoretical time Publics de la Région Lausannoise), ATM of arrival. Regularity can be calculated as (Azienda Transporti Milanesi), BCRTC the percentage of trains respecting the (British Colombia Rapid Transit), GTT theoretical time of arrival. (Gruppo Torinese di Transporte), TMB (Transports Metropolitans de Barcelona), Regularity: an indicator quantifying the TRTC (Taipei Rapid Transit). difference between actual headways and theoretical headways. Regularity can be Light rail transit (LRT): a system operat- calculated as the percentage of trains ing on a circuit that may be shared with respecting theoretical headways. other modes of transport (such as cars), often powered by a catenary, the trains Urbanization: population movements from of which have a limited number of cars (1 rural areas to urban areas, or the transfor- or 2). Trams are included in this category. mation of rural areas into densely populated urban areas. Metro: a mass transit system operating on a dedicated underground or overground circuit. Generally speaking, the train is made up of several cars (up to 6 or 8) and comprises a raised access level.

102 Frame of references

“The trend to automation”, London Seminar, UITP 2013

“Metro automation – facts and figures”, UITP

“Observatory of automated metros 2013”

“UTO Metros opportunities and challenges”, Arts et métiers

“The automation control systems for the efficiency of metro transit lines”

“Impacts of Unattended Train Operations on Productivity and Efficiency in Metropolitan Railways”, Judith M Cohen, Alexander S Barron, Richard J Anderson, Daniel J Graham, 2015, Transportation Research Board

“The impacts of moving block signaling on technical efficiency: An application of pro- pensity score matching on urban metro rail Systems”, Shane Canavana, Daniel J. Graham, Patricia C. Melo, Richard J. Anderson, Alexander Barron and Judith M. Cohen, 2015, Railway and Transport Strategy Centre, Centre for Transport Studies, Dept of Civil and Environmental Engineering, Imperial College London

“Future of rail 2050”, ARUP

“Copenhaguen metro – The best metro in the world 2010”, Ansaldo STS

“The world cities in 2016”, Nations Unies

“The economic of public transportation 2007”, Taylor & Francis

“The future of Transportation 2015”, Siemens

2015 Annual activity reports for metro operators and suppliers

INSEE communications

LVMT Laboratoire Ville, Mobilité et Transport communications

Specialised press articles: Streetsblog, CityLab, Nordbayern, Railway Technology, Gulf News, Think Railways, MarktSpiegel, Business Insider…

103 www.wavestone.com 104